Learning with the machines

A man with glasses works on a laptop between two piles of books, one with a robot on top and one with a coffee cup on top.
Graphic by Victor Bonhart.

In a recent TikTok video, Chris Kanich, associate professor of computer science at UIC, took a provocative stance on what AI means for today’s college students, who are the first to go through school knowing that AI could do their homework for them. 

“This Wild West of having large language models everywhere is going to be a whole lot of fun, and we’re going to try and figure out how to deal with it,” he says in his video, which attracted over 300,000 views and 1,000 comments in the first week after posting. 

Large language models, the AI behind applications such as OpenAI’s ChatGPT and Google’s Bard, abruptly broke through in late 2022. Built on massive volumes of data and the artificial intelligence approach known as deep learning, these models produce realistic text and conversations based on any user prompt. Ask them to write an essay about bird migration, a cover letter for a job application, or computer code for an interactive graphic, and the response will be eerily human – if often slightly inaccurate. 

Across the University of Illinois Chicago, scholars wasted no time exploring what this technology means for research and education. From the English department and the Center for the Advancement of Teaching Excellence to the College of Engineering and UI Health, people are experimenting with these tools’ generative text and imagining how they may transform their field and the next generation of students and workers.  

“I feel like we’ve never faced a challenge quite like AI,” said Mark Bennett, clinical associate professor of English and director of the First-Year Writing Program. “It’s compelling us to rethink every aspect of our teaching.” 

But alongside the challenges come opportunities in every corner of higher education. At UIC, physicians used ChatGPT to write a research paper, students dissected its advantages and shortcomings in first-year writing courses, education experts created guides for using the tool in the classroom, and computer scientists imagined what comes next in language models. 

“My fundamental interest and concern is not so much what makes it tick, but how it’s going to evolve,” Kanich said. “From a computer science perspective, we have a reasonable understanding of what it’s good at and what it’s not. The more interesting thing is going to be how people end up using it and what it gets used for, because there is a lot of creativity yet to be expressed.” 

Read more about how UIC educators and researchers are exploring the impact of ChatGPT and language models in the stories below:

AI in the classroom: opportunity instead of fear 

Rapidly adapting to AI in education and the library

Medical writing with language models: faster but faultier

Improving the models – smaller data, bigger dreams

Print Friendly, PDF & Email


Research, UIC today